Re: Why we lost Uber as a user

On Wed, Aug 3, 2016 at 3:30 AM, Alfred Perlstein <alfred(at)freebsd(dot)org> wrote:
> We may be saying the same thing, but still there is something to be said for
> logical replication... also, didnt they show that logical replication was
> faster for some use cases at Uber?

There is certainly something to be said for logical replication just
as there is something to be said for having regular pg_dumps which are
logical exports of your database. But neither is a substitute for
having real backups or a real standby database. They serve different
purposes and solve different problems.

But when you have a hardware failure or physical disaster the last
thing you want to be doing is failing over to a different database
that may or may not have the same data or same behaviour as your
former primary. You want to switch over to a standby that is as near
as possibly byte for byte identical and will behave exactly the same.
If there was a bug in your primary the last time you want to find out
about it and have to be dealing with fixing it is when you have a
disaster in your primary and need to be back up asap.

Honestly the take-away I see in the Uber story is that they apparently
had nobody on staff that was on -hackers or apparently even -general
and tried to go it alone rather than involve experts from outside
their company. As a result they misdiagnosed their problems based on
prejudices seeing what they expected to see rather than what the real
problem was.

--
greg